- Custom architecture
- Purpose-built development
- Systemic view
AI and data systems that actually work, not just in a slide deck
I help companies turn an AI idea into a working system: architecture, agents on local or cloud models, databases and ETL, and automation that saves real hours.
- Direct, personal contact
- Defense and enterprise background
- Hebrew and English
- [AI Agent] processing data...
- [Webhook] triggered successfully
- [Data Pipeline] syncing 4 APIs...
- [Deployment] optimization complete
- RAFAEL, Iron Dome
- Israel Aerospace Industries
- Israeli Air Force
- Landa
- IMI Systems
- Enterprise & media clients
A systemic view, from data to outcome
Every project starts with the full picture: what data exists, how it flows and what has to happen at the end. Then we build only what's needed.
01Systems engineering & API integration
Custom architecture for the whole system: how data flows, which systems talk to each other and where AI actually adds value. CRM, ERP and internal systems connected through stable, monitored interfaces.
Architecture · REST · Webhooks · SSO02Purpose-built AI components
Agents, RAG and semantic search over your documents, on local models for privacy or cloud models where they fit.
Agents · RAG · Ollama · MCP03Business process automation
Quotes, tagging, meeting summaries and reports that happen on their own, with human review where it matters.
RPA · n8n · Workflows04Databases & ETL
Database design, ETL pipelines, and cleaning and unifying data from many systems into one reliable source.
PostgreSQL · Oracle · Qdrant05Cloud & infrastructure
Set up or migrate to AWS and GCP, or run on-prem and hybrid: containers, security, SSO and monitoring.
AWS · GCP · Docker · Grafana06Consulting & assessment
Build vs. buy, stack selection and an architecture document with phases, costs and risks before you commit to development.
Assessment · Roadmap · Due Diligence07Fractional CTO
Senior technology leadership part-time: build the team, set engineering process, and support you with investors and clients.
Fractional CTO · Team · Process
Three focused steps
01Intro call
A short phone or Zoom call: the challenge, the systems and data you have, and what success looks like.
Phone or Zoom02Assessment & plan
An architecture document and a work plan with phases, costs and risks, so you know exactly what you're deciding on.
Document & work plan03Build & deploy
Short iterations with working deliverables, deployment in your environment, monitoring and handover to your team.
Short, working iterations
It can start with one conversation.
Systems running at clients today
- Enterprise AI, B2B
QuoteAI Pro – AI proposal platform
A 9-stage workflow with AI at every step, RAG over client documents, a local-and-cloud model chain and Hebrew speech-to-text.
Proposal cycle cut from days to hours
- Media
TTags – smart media asset management
Automatic tagging of video and images with semantic search across the whole library and secure multi-team access.
Hours of daily manual tagging eliminated
- Productivity
GIGI – meeting intelligence
Recording, multi-speaker transcription in Hebrew and English, and automatic summaries and action items.
Used daily by management and ops teams
- Mission-critical
Signal/CELI – emergency comms
A real-time SMS command dashboard integrated with national telecom infrastructure for emergency coordination.
Continuous operation under zero-downtime requirements
Yaron Elgrabli, CTO & Systems Architect
I started out building systems that are not allowed to fail: 12 years at RAFAEL, including Iron Dome and the Skylite micro-UAV. Then systems architecture at IAI, IMI and Landa, and co-founder and CTO of a startup.
Today I build and operate AI systems for clients myself, so every recommendation I make is grounded in what actually works.
- 25+years building and leading systems
- 12years at RAFAEL, incl. Iron Dome
- 10+AI products live with clients
- 87+automation workflows in production
Before you call
Who is this for?
Companies that want AI and automation in their processes, startups that need technology leadership, and teams that need an architect for a complex project.
Can models run without sending data to the cloud?
Yes. I specialize in running local models on your own infrastructure, and add cloud models only where they're justified.
How do we start?
Book a short consultation call. By the end you'll know whether there's a fit and what the next step is, with no commitment.
Do you work with an existing dev team?
Absolutely. I can work alongside your team as architect and technical lead, or deliver an end-to-end solution.
Shall we book a consultation?
Tell me briefly what you need and I'll get back to you to set a time. Fastest option: just call.
054-992-6200- AWS
- Google Cloud
- PostgreSQL
- Oracle
- Qdrant
- n8n
- Docker
- Ollama
- Grafana
- MCP
Names refer to technologies I use in projects, not official partnerships.